Dairy production evolution in an export-oriented country.

Uruguay is a small market with a large export-oriented agribusiness sector. That imposes a challenge for being competitive on the international markets. One of the sectors that is most affected by the international competition is dairy production, where subsidies and market power of large enterprises rules the market. Its main export products are commodities, making it contingent on international prices. Over the years the dairy systems have shifted from an extensive grass-fed model into more intensive farming combining cultivated pastures and larger amounts of high-quality feed. Production saw a maximum in 2013 and a decline since 2014. Another side of this intensification is that as production increases, dairy farms decreases. Between 2011 and 2020 the number of dairy farms dropped 25% while the daily milk production per cow increased more than 50%. Comparing the results of the surveys it seems clear that there are farms running out of business from all sizes, so it is not only a scale issue. This put a question on whether the intensification is sustainable or not and challenge the sector on what are the causes that push farms out of business. To assess these questions, we propose a technical efficiency analysis using two national dairy farm surveys conducted by INALE (Instituto Nacional de Lechería) in 2014 and 2019 using a meta frontier approach to compare the change over time since the surveys are independent and do not compose a panel. The meta frontier allows to compare each survey as a unique group but as of using a technological common source for which a Propensity Score Match is necessary to address biases from observables. From this estimation we can find technical efficiency scores by survey and meta technical efficiency scores that reports the distance to the meta frontier. From that result we can infer if the yield improvement from one survey year to the next is due to movement towards the frontier (improved efficiency) or a productivity improvement due to a shift in the frontier. From the results we expect to provide technology paths that show better performance as possible extension and policy leading recommendations. At this point we are working on the database set up to perform the estimation, so we do not have preliminary results